How That Post Was Made
February 3, 2026
Showing the receipts
These posts are written with AI assistance. If you've been reading this series, you probably suspected as much. A writer arguing that "specification is the new bottleneck, that implementation is being commoditized" would naturally be using these tools himself.
This is how "The Specification Is the Skill" was created. Not as a defense of the practice, but simply showing the process.
The Conversation That Started It
After publishing "When the Code Becomes Optional," the author brought it to GPT for feedback. Rather than receiving edits, he got something generative—GPT identified an unstated core idea: "The implementation cost has collapsed to near-zero, but the specification cost has not."
GPT pushed further: "The new senior engineer skill isn't 'knows Rust' or 'knows Kubernetes.' It's 'can fully specify intent under uncertainty.'"
Recognizing this had become another post entirely, the author asked GPT to write a draft specification, deliberately demonstrating the thesis itself.
The Spec
GPT produced a complete specification document covering purpose, audience, thesis, non-goals, required sections with acceptance criteria, tone constraints, and success metrics.
Core thesis: "Specification is no longer documentation—it is the primary act of engineering."
Non-goals included: avoiding claims that AI will replace programmers, dismissing model limitations, or claiming this applies equally to all engineering work.
Acceptance tests: Would a senior engineer say this explains frustrating AI experiments? Would a manager understand why only certain team members gain leverage?
The non-goals proved most useful by constraining sprawl before it started.
The Handoff
The author deliberately brought the specification to Claude in a separate conversation to test portability. Claude flagged structural concerns and suggested reframing: specification wasn't emerging as a new skill—"it's an existing skill becoming visible and therefore trainable."
The specification proved portable, independent of the model that created it.
Where Specs Come From
An important clarification: specifications don't emerge from single prompts. Before requesting the spec, there was lengthy conversation—research, back-and-forth exploration, rabbit holes. The spec represented crystallization of already-completed work.
The Draft and Editorial Pass
Claude generated a complete draft matching the specification structure in roughly ninety seconds, flagging uncertainties about examples and provocative framing choices.
The editorial pass was minimal. The author caught modest overreach but found the structure sound—demonstration that robust specifications enable lighter intervention.
The Authorship Question
When asked directly about authorship, Claude responded: whoever "can fully specify intent under uncertainty." The author generated the text; the specification writer determined value.
Claude offered an analogy: "You're doing something closer to what a film director does." The implementation—like operating a camera—remains commoditized. The leverage points stay human.
The Recursion
Creating this meta-post required its own specification, written by Claude, describing how GPT had specified the previous post that Claude executed—demonstrating the thesis through its own production process.
What This Proves (and Doesn't)
This process works for analytical, structured content with clear theses. It fits arguments and technical explanations where shape can be predetermined.
It wouldn't work for poetry or personal essays—though one might specify the decision about whether to write them.
The remaining human leverage points: initial direction, judgment about specification sufficiency, model selection, verification against overreach. These haven't been commoditized yet.
Appendix: The Specs
Specification for "The Specification Is the Skill"
Purpose: Argue that specification—not implementation—has become the primary engineering skill.
Target Audience: Senior engineers, staff/principal ICs, managers, technical executives.
Core Thesis: "Specification is no longer documentation—it is the primary act of engineering."
Non-Goals: Predicting AI timelines, declaring specialists obsolete, advocating specific tools.
Tone: Calm, precise, non-evangelical. Avoid revolutionary language.
Success indicators: Senior engineers recognize their AI frustrations. Managers understand leverage distribution. Skeptics acknowledge substance despite disagreement.
Specification for "How That Post Was Made"
Purpose: Demonstrate the thesis by transparently showing how the previous post was produced.
Core Thesis: The production process itself evidences the argument about specification as primary skill.
Non-Goals: Defending AI writing's legitimacy, providing prompting tutorials, overselling output quality, hiding rough edges.
Tone: Matter-of-fact documentation. Show, don't defend.
Success indicators: Skeptics acknowledge transparency. Practitioners could replicate this workflow. Readers feel they've seen behind the curtain.
This is a follow-up to The Specification Is the Skill.